During the 1956 Dartmouth Summer Research Project on Artificial Intelligence, John McCarthy developed early important concepts for Lisp. But, the idea of the Lisp machine was in the late 1960s. Later in 1969, Anthony Hearn and Martin Griswold of the University of Utah established Standard Lisp, the first attempt at Lisp standardization.

Likewise, in the early 1970s, Peter Deutsch and Daniel Bobrow used microcode to parse a byte-code implementation language on the Alto, a single-user minicomputer. Finally, at MIT, Richard Greenblatt began developing new hardware and instruction set architecture.

Why Lisp?

In the 1960s and 1970s, artificial intelligence (AI) computer programs needed a lot of processor time and memory space, which was a lot of computer power. The Lisp symbolic programming language made AI research even more power-hungry when commercial hardware was designed and optimized for programming languages like assembly and Fortran. The name of the first machine was CONS (named after the list construction operator cons in Lisp). Later, object-oriented programming ideas significantly affected Lisp in the late 1970s.

Furthermore, lisp machines commercially pioneered numerous now-common technologies, including efficient garbage collection, laser printing, windowing systems, computer mouse, high-resolution bit-mapped raster graphics, computer graphic processing, and networking breakthroughs like Chaosnet.

When did it start?

At first, the Lisp machines were personal workstations for writing Lisp software. Initially, there was no way for more than one person to use them at once. Likewise, the machines had a large black-and-white bitmap display, a keyboard and mouse, a network adapter, local hard disks, more than 1 MB of RAM, serial interfaces, and a local bus for expansion cards. There was no need for colour graphics cards, tape drives, or laser printers.

The processor did not run Lisp directly. Instead, it was a stack machine with instructions that researchers optimized for compiled Lisp. The instruction set for the first Lisp machines was the microcode. During runtime, type checking and dispatching were in hardware for several operations. For example, researchers could only add different types of numbers with one function (integer, float, rational, and complex numbers). The result was a tiny version of Lisp code that researchers had compiled.

AI and Lisp

McCarthy's work led to a new field of study in the middle to late 1960s, where AI, machine learning (ML), and deep learning became possible. And Lisp quickly became the standard in this new area. People say that McCarthy bet a Scottish chess master, David Levy, in 1968 that a computer would be able to beat Levy in a chess match in 10 years. 

Important lisp milestones are:

• Researchers implemented early varieties of Lisp on the IBM 704, the IBM 7090, the Digital Equipment Corporation (DEC) PDP-1, the DEC PDP-6, and the PDP-10. 
• Lisp 1.5 was the most widespread dialect of Lisp between 1960 and 1965. 
• MacLisp and Interlisp were the two prominent dialects of Lisp by the early 1970s, both of which originated from these early attempts.
• Although the first implementations of Lisp were on the IBM 704 and the IBM 7090, later work focused on the DEC PDP-6. 
• The latter is the mainstay of Lisp and artificial intelligence work at MIT, Stanford University, and CMU from the mid-1960s through the 1970s.

Conclusion

Researchers used Lisp to write all of the software for operating systems. When Common Lisp came out, researchers supported Common Lisp on the Lisp Machines, and some system software was ported to Common Lisp or later written in Common Lisp.

Likewise, some of the later Lisp machines, such as 

• the TI MicroExplorer, 
• the Symbolics MacIvory, and 
• the Symbolics UX400/1200. 

The above lisp machines were no longer complete workstations. Instead, they were boards that researchers could put into host computers, such as the Apple Macintosh II and the SUN 3 or 4. 

Moreover, some Lisp machines, like the Symbolics XL1200, could do a lot of graphics work with the help of unique boards. For example, researchers used these machines in 3D animation, CAD, and medical image processing.